Hi Nikolaus,
On Wed, 19 Sep 2018 12:57:34 +0200 Andreas Kemnade andreas@kemnade.info wrote:
On Wed, 19 Sep 2018 11:18:18 +0200 "H. Nikolaus Schaller" hns@goldelico.com wrote:
Hi Andreas,
Am 18.09.2018 um 21:34 schrieb Andreas Kemnade andreas@kemnade.info:
Hi,
On Tue, 18 Sep 2018 08:30:58 +0200 H. Nikolaus Schaller hns@goldelico.com wrote:
Hi all, I have finally updated our kernel trees to
letux-4.14.70 letux-4.18.8 letux-4.19-rc4
and as usual there are binaries and sources at the known locations.
Regarding 4.19-rc4 we have quite some construction sites, unfortunately with some regressions.
Here is the list of (significantly) broken things that I am aware of:
Pyra:
- OMAP5 TILER is no longer scanning through the frame buffer and appears to fetch from a fixed address
- "reboot" hangs if timer8 is in use
- bq2429x charger driver is far from being perfect (sometimes does not detect USB unplug/replug and charging does not start)
- AESS does not load firmware and even without, audio is broken
- PVR/SGX integration is missing
- sd/eMMC switch driver is missing
GTA04:
- camera driver is still not stable and incomplete
- PVR/SGX integration is incomplete
suspend works catastrophically at least on GTA04A5, totally unreliably. Do not know yet why.
Well, I might have mixed up something during manually fixing rebase conflicts...
Can you be a little more specific?
Well, I am still collecting and verifying (double checking things I have done). What I have seen:
- 10mA too much power consumption if no modules loaded
- is it not going to suspend?
per and core power domains do not enter suspend state noirq suspend failed
on letux3704 that happens after loading camera modules, but also after other things I have not figured out. Running udev is enough. musb + phy + charging stuff considered not gulthy at first glance.
ok, checked more cleanly. Suspend on letux3704 seems not to be problematic, if omap3_isp is blacklisted, so maybe something there has to learn to clean up on probe failure. So either omap3_isp or the ov9650 module. GTA04A5 is a bit different story. Suspend problems there seems to be present starting at 4.19-rc1 even on mainline. But I will double-check that. At that evening I had just the idea to produce a quick v2 of my phy patch where Dmitry had some objections, so I was just searching for a good base to develop and test a v2.
Well, worst case will be an automated git bisect run next night. I have now a nice sdcard to jst-xh adapter "battery". So all what is additionally needed is one script which returns 0 if kernel is ok, 125 if it cannot be tested, something else on failure. And then I can do a git bisect run script.sh
Regards, Andreas